Navigating New Spaces: A Better Way to Train Agents
Imagine an agent that can understand your instructions and navigate through various environments. This is what Vision-and-Language Navigation (VLN) aims for. However, training these agents to handle new, unseen spaces is a challenge. Why? Well, there's limited data and the environments these agents